REVIEW 3 major objections 5 minor 16 references
Stochastic birhythmicity outside the coexistence region in the Hindmarsh-Rose model
T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash
Pith's one-line read Noise extends neuron birhythmicity: the Hindmarsh-Rose model switches between two bursting rhythms even outside the deterministic coexistence interval.
desk verdict Noise-induced birhythmicity beyond deterministic coexistence in the Hindmarsh-Rose model is credible and the analytic toy model is clean; the quantitative HR phase diagram is cutoff-dependent, but that soft spot does not sink the main claim. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing object is the ghost of a limit cycle near a saddle-node bifurcation: a transiently attracting, long-lived orbit that exists on the side of the bifurcation where the periodic solution has disappeared. In the Hindmarsh-Rose model the ghost is triggered by additive noise on the slow variable $z$, and the paper detects it through the $z$-amplitude, whose value separates two-spike bursts (below 0.9) from three-spike bursts (above 0.9). The analytical backbone is a simple axially symmetric model with radial dynamics $f(r) = -r\left((r^2-1)^2-b\right)$ and a constant rotation term; for this model the stationary Fokker-Planck equation has the exact solution $\rho_s(r) = C \exp(-2u(r)/\varepsilon^2)$ with $u(r) = \tfrac12\left[\tfrac13(r^2-1)^3 - b r^2\right]$, so stochastic equilibrium states are the extrema of the radial distribution, given by $b(r) = -\varepsilon^2/(2r^2)+(r^2-1)^2$. The folds of this branch are stochastic saddle-node bifurcations, parameterized by $b=(1-r^2)(1-3r^2)$ and $\varepsilon = 2r^2\sqrt{1-r^2}$, and they merge in a cusp at $(b,\varepsilon)=(-1/3,\,4/(3\sqrt{3}))$, the analogue of a stochastic pitchfork.
What would settle it
Run the Hindmarsh-Rose model at $b=2.906$ with $\varepsilon$ in the range $1\times 10^{-3}$ to $6\times 10^{-3}$ for at least $10^6$ bursting oscillations and count how often a 2-cycle (with $z$-amplitude below 0.9) occurs: if the proportion stays below the 2% cutoff for all $\varepsilon$ below the claimed crossing, the extension of the birhythmicity region is a finite-time artifact. For the simple model, measure the radial distribution at $\varepsilon=0.9$: since 0.9 exceeds the cusp value $4/(3\sqrt{3})\approx 0.77$, theory predicts a single maximum everywhere, so observing two well-separated peaks for any $b$ would refute the analytic saddle-node curves.
Extended reading notes
Core claim
On the paper's own terms, the central discovery is that stochastic birhythmicity is not confined to the deterministic bistable interval $2.9082 \le b \le 2.9231$ of the Hindmarsh-Rose model. For $b$ just outside this interval, where only the 2-cycle or only the 3-cycle is a stable bursting attractor, sufficiently strong additive noise induces recurrent transitions between that attractor and the ghost of the other bursting cycle, so the system spends time in both rhythms and the histogram of $z$-amplitudes shows two distinct peaks. The paper therefore treats the noise as extending the birhythmicity region, with the left boundary crossing the deterministic saddle-node near $\varepsilon \approx 2\times 10^{-3}$ and the right boundary near $\varepsilon \approx 3\times 10^{-3}$. The same qualitative behavior is established in a simple two-variable model with a radial potential and a single saddle-node bifurcation; there, stationary probability maxima are computed from the exact Fokker-Planck solution and the boundaries of stochastic birhythmicity are given analytically as a parametric curve ending in a cusp.
Load-bearing premise
The paper assumes that the numerically estimated boundaries of stochastic birhythmicity — based on a 2% occurrence cutoff over about 20,000 oscillations and a $z$-amplitude threshold of 0.9 — represent the asymptotic stationary behavior; the simulations in the simple model show that finite observation time can hide rare states, so the Hindmarsh-Rose boundaries could shift with longer runs or a different cutoff.
Editorial extensions
If this is right
- In the Hindmarsh-Rose model there is a de facto stochastic birhythmicity region in the $(b,\varepsilon)$ plane that is wider than the deterministic one once $\varepsilon$ exceeds roughly $2\times 10^{-3}$ on the left and $3\times 10^{-3}$ on the right; for large noise, both bursting rhythms occur with comparable probability.
- The stochastic birhythmicity region narrows before it widens as noise increases: near $\varepsilon \approx 10^{-3}$ it shrinks to about a third of the deterministic interval because transitions preferentially fall into the lower-energy 2-cycle state.
- In the simple model, noise creates a large-orbit stochastic state for $b<0$, beyond the deterministic saddle-node bifurcation, and the stochastic saddle-node branches delimit a pointy birhythmicity region with a cusp; above the cusp noise the two stochastic states merge into one.
- Since the effect depends only on saddle-node ghosts, any neuron model whose bursting branch folds at a saddle-node bifurcation should show analogous noise-driven two-rhythm switching near the fold.
Reading between the lines
- Beyond what the paper states, the same ghost-based mechanism should operate near every saddle-node fold in the bursting branch, not just the 2-cycle/3-cycle fold studied here; other spike-count transitions in the Hindmarsh-Rose parameter space should exhibit similar noise-driven switching.
- The simple model's exact stationary distribution suggests a quantitative scaling: the width of the stochastic birhythmic region in the control parameter should grow roughly like a power of $\varepsilon$ near each deterministic fold, so measuring that width in the Hindmarsh-Rose model could test whether a universal shape law holds.
- Biologically, this implies that channel-noise variability could let a neuron express a rhythm not encoded in its deterministic parameters, effectively adding an extra functional mode that could be either computational flexibility or a source of variability, depending on the context.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies the Hindmarsh-Rose (HR) neuron model with additive white noise on the slow variable z. In the deterministic case, the model has two coexisting bursting orbits (2-cycle and 3-cycle) in a narrow interval of the control parameter b between two saddle-node bifurcations of limit cycles (about b=2.9082 and 2.9231). The authors report that for b outside this interval but close to it, noise induces random switching between the unique stable bursting attractor and the ghost of the other attractor, so that the effective birhythmic region is extended by noise. They map this stochastic birhythmicity region in the (b,ε) plane using numerical simulations and a 2% occurrence cutoff. To put this on an analytical footing, they introduce a simple axially symmetric two-variable model with a periodic-solution saddle-node bifurcation, solve the stationary Fokker-Planck equation, derive the stochastic equilibrium branches and the stochastic saddle-node boundaries, and locate the cusp where the boundaries meet at (b,ε)=(-1/3,4/(3√3)). They show that in the toy model the stochastic birhythmicity region extends beyond the deterministic saddle-node, but not beyond the Hopf bifurcation, analogous to the HR result.
Significance. The analytical part is a genuine strength: the stationary Fokker-Planck solution, the branch equations (Eqs. (11)-(14)), and the cusp calculation are derived without fitting parameters, giving a clean demonstration that an additive-noise-driven system can have two stable stochastic equilibrium states beyond a saddle-node bifurcation, with the boundaries computable in closed form. The HR observation is qualitatively plausible and well illustrated by the time series and histograms (Fig. 3) and by the deterministic ghost simulations (Fig. 4). If the quantitative phase diagram were reliable, the paper would provide a useful map of noise-induced birhythmicity in a widely used neuronal model. However, the HR phase diagram in Fig. 6 rests on an arbitrary 2% cutoff and finite-time simulations, which undermines the precision of the reported boundaries, though not the qualitative phenomenon. The toy model cannot be used to calibrate the HR cutoff, so the HR claims need additional support to be fully convincing.
major comments (3)
- [Section II, Fig. 6 and following text] The stochastic birhythmicity region in the HR model is defined by a 2% occurrence cutoff over about 20,000 oscillations per parameter point, a threshold that the authors themselves call 'rather arbitrary.' The deterministic ghost has a long, parameter-dependent lifetime (Fig. 4), so finite-time simulations can show both z-amplitude classes even if the stationary distribution is unimodal. The paper's own discussion in Sec. III ('Longer simulations would allow for the less likely stable states to be eventually observed') shows that finite-time statistics underrepresent rare states. Consequently, the reported boundaries—including the narrowing before widening and the crossing points at approximately ε=2×10^-3 and 3×10^-3—are not established as stationary properties and may shift with the cutoff, the simulation length, and the initial-condition protocol. The authors should either base the boundaries on an estimated stationary distribution (with convergence checks) or show a sensitivity analysis of Fig. 6 with respect to these choices, and temper the claim that the noise 'effectively extends the birhythmicity region' to reflect the operational nature of the definition.
- [Section II, caption of Fig. 3 and text following Fig. 5] The classification of bursts into 2-cycle and 3-cycle uses a fixed z-amplitude threshold of 0.9. The z-amplitude distributions broaden with increasing noise (Fig. 5), and for large ε the amplitude ranges of the two classes appear to overlap, as the authors note that amplitudes 'fluctuate greatly from their average values.' This raises the possibility that the 2% proportions in Fig. 6 are sensitive to the chosen threshold. The authors should either count spikes directly to classify bursts or demonstrate that the phase diagram in Fig. 6 is robust to reasonable variation of the threshold.
- [Section II, paragraph after Fig. 4] The sentence 'In all cases, the system exhibits stochastic birhythmicity, without distinguishing between actual attractors and their ghosts' blends two distinct notions. In the toy model, stochastic birhythmicity is rigorously defined through the bimodality of the stationary Fokker-Planck distribution. In the HR model, the evidence outside the coexistence region consists of finite-time observations of two amplitude classes, which can also arise from long transients associated with a ghost in a system with a single attractor. The paper should clarify whether stochastic birhythmicity in the HR model is intended as a stationary, bimodal property (and if so, provide evidence from long-time stationary histograms) or as a finite-time switching phenomenon (and if so, avoid the implication of two coexisting stable states).
minor comments (5)
- [Sec. III, Eq. (3)] The symbol r denotes both the two-dimensional vector and its Euclidean norm; please use a distinct symbol for the radial coordinate (e.g., ρ) to avoid ambiguity.
- [Sec. III, text after Eq. (7)] The phrase 'replaced by an new notion' should be 'replaced by a new notion.'
- [Sec. III, below Eq. (14)] The term 'sadle-node' is a typo for 'saddle-node.'
- [Figure 10 caption] The caption states that the saddle-node branches emerge from the deterministic saddle-node and Hopf bifurcations at (b,ε)=(0,0) and (1,0); please label which branch corresponds to each deterministic bifurcation origin, since both branches meet at the cusp.
- [Title page, PACS numbers] The PACS numbers listed (02.02.30.Oz, 02.02.50.Ey, 05.05.10.Gg) appear malformed; they should be checked against the standard PACS scheme.
Circularity Check
No significant circularity: HR stochastic-birhythmicity claims are direct numerical observations, and the toy-model analysis is an independent Fokker-Planck derivation.
full rationale
The paper's central HR claim—that noise induces two-state bursting switching outside the deterministic bistability interval—is supported by direct numerical simulations of Eqs. (1–2), with the two bursting states classified by a stated z-amplitude threshold and a 2% occurrence cutoff. No parameter is fitted to reproduce the target result; the boundaries in Fig. 6 are measured from the model dynamics themselves. The toy model in Sec. III is an independent construction: its radial potential and additive noise are specified a priori, and the analytical stochastic equilibrium branches, Eqs. (11)–(14), are derived from the associated Fokker-Planck equation rather than imported from the HR simulations. The cited Refs. [6] and [13] supply parameter conventions and a standard stationary Fokker-Planck solution, but the paper explicitly re-derives the latter and the HR observations stand on their own. The acknowledged limitations concerning the arbitrary 2% cutoff and finite simulation time are interpretive and statistical issues, not circular reasoning.
Assumptions & free parameters
free parameters (3)
- z-amplitude classification threshold =
0.9
- stochastic birhythmicity proportion cutoff =
2%
- observation length per parameter point =
about 20000 oscillations
assumptions (5)
- standard math The Fokker-Planck equation (6) describes the stochastic process (3)-(5) with additive white noise in the Itô-Stratonovich sense.
- standard math The stationary density is obtained by setting the probability current to zero and assuming no probability flux at infinity (Eqs. 7-8).
- domain assumption The deterministic HR parameter values (a=1, c=1, d=5, s=4, x0=-1.6, r=0.01, I=2.2) are taken from Ref [6] and reproduce pond snail neuron behavior.
- domain assumption For the HR model, noise is applied only to the slow variable z because random fluctuations in z are 'what essentially leads to transitions between stable attractors' (Section II).
- domain assumption The deterministic HR model has exactly two stable bursting cycles (2-cycle and 3-cycle) with saddle-node bifurcations at b=2.9082 and b=2.9231 for the chosen parameters.
Cite this review
Pith. "Pith review of Stochastic birhythmicity outside the coexistence region in the Hindmarsh-Rose model." pith.science (2026). https://pith.science/paper/6HSJPHHG
@misc{pith2026250505972,
author = {Pith},
title = {Pith review of: Stochastic birhythmicity outside the coexistence region in the Hindmarsh-Rose model},
year = {2026},
howpublished = {\url{https://pith.science/paper/6HSJPHHG}},
note = {Machine review of arXiv:2505.05972}
}
read the original abstract
In this work, we demonstrate that the Hindmarsh-Rose model subjected to additive white noise exhibits birhythmicity. Specifically, the system fluctuates between two distinct bursting attractors characterized by different numbers of spikes. This behavior is observed not only within the bistable region bounded by two saddle-node bifurcations of limit cycles but also beyond these boundaries. This phenomenon is associated with the ghost effect, typically observed near deterministic saddle-node bifurcations. We map the region of stochastic birhythmicity in terms of the noise intensity and a key deterministic parameter that controls the dynamics of fast ion channels. To provide an analytical foundation, we introduce a simple stochastic model with a single saddle-node bifurcation. In this model, stochastic birhythmicity is similarly characterized as a function of noise intensity and the control parameter.
Figures
Figures from the paper (7 more)
Reference graph
Works this paper leans on
-
[1]
Izhikevich, Dynamical systems in neuroscience: the geometry of excitability and bursting
E. Izhikevich, Dynamical systems in neuroscience: the geometry of excitability and bursting . (MIT Press, Cambridge, 2007)
work page 2007
-
[2]
A. L. Hodgkin and A. F. Huxley, A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol. 117, 500 (1952)
work page 1952
-
[3]
J. L. Hindmarsh and R. M. Rose, A model of neuronal bursting using three coupled first order differential equations. Proc R Soc Lond B Biol Sci 221, 87 (1984)
work page 1984
-
[4]
A. Goldbeter and J. Yang, Multi-synchronization and other patterns of multi-rhythmicity in oscillatory biological systems. Interface Focus. 12, 20210089 (2022)
work page 2022
-
[5]
E. Njinkeu Nganso, R. Mbakob Yonkeu, G. Filatrella, and R. Yamapi, Multi-rhythmic oscil- lations and correlated noise effects of a self-sustaining biological system. Nonlinear Dyn. 108, 4315 (2022)
work page 2022
-
[6]
E. Slepukhina, I. Bashkirtseva, P. K¨ ugler, and L. Ryashko. Noise-driven bursting birhythmicity in the Hindmarsh–Rose neuron model. Chaos 33, 033106 (2023)
work page 2023
- [7]
-
[8]
S. Strogatz, Nonlinear dynamics and chaos: with applications to physics, biology, chemistry, and engineering. (Westview Press, Boulder, 2015)
work page 2015
Show all 16 references
-
[9]
Helmholtz, ¨Uber Integrale der hydrodynamischen Gleichungen, welche den Wirbelbewe- gungen entsprechen
H. Helmholtz, ¨Uber Integrale der hydrodynamischen Gleichungen, welche den Wirbelbewe- gungen entsprechen. J. Reine Angew. Math 55, 25 (1858), 21
-
[10]
Gardiner
C. Gardiner. Handbook of stochastic methods for physics, chemistry and the natural sciences . (Springer, Berlin, 2004)
2004
-
[11]
G. Agez, M. Clerc, and E. Louvergneaux, Universal shape law of stochastic supercritical bifurcations: Theory and experiments. Phys. Rev. E 77, 026218 (2008)
2008
-
[12]
G. Agez, M. Clerc, E. Louvergneaux, and R. Rojas, Bifurcations of emerging patterns in the presence of additive noise. Phys. Rev. E 87, 042919 (2013)
2013
-
[13]
Ortega, M
I. Ortega, M. Clerc, C. Falc´ on, and N. Mujica, Subharmonic wave transition in a quasi-one- dimensional noisy fluidized shallow granular bed. Phys. Rev. E 81, 046208 (2010)
2010
-
[14]
Fokker, A
A. Fokker, A. Die mittlere Energie rotierender elektrischer Dipole im Strahlungsfeld. AdP 348, 810 (1914),
1914
-
[15]
Planck, ¨Uber Einen Satz Der Statistischen Dynamik Und Seine Erweiterung In Der Quan- tentheorie
V. Planck, ¨Uber Einen Satz Der Statistischen Dynamik Und Seine Erweiterung In Der Quan- tentheorie. (Sitzungsberichte der K¨ oniglich Preussischen Akademie der Wissenschaften zu Berlin, Berlin, 1917)
1917
-
[16]
Kolmogorov, ¨Uber die analytischen Methoden in der Wahrscheinlichkeitsrechnung
A. Kolmogorov, ¨Uber die analytischen Methoden in der Wahrscheinlichkeitsrechnung. Math. Ann. 104, 415 (1931). 22
1931
Reviewed August 15, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.